Main Analysis

Proportion of Cats and Dogs Prealtered

PetPoint includes a field called prealtered (Whether an animal was already altered when entering the shelter). Based on our sample, the proportion of cats and dogs entering the shelter prealtered progressively decreased from 2019 - 2023. The percentage point decrease from 2019 to 2023 for dogs (10.92) was larger than for cats (5.9). It is important to note that the proportion of prealtered cats and dogs entering the shelter in 2023 is roughly the same, 21.95 (cats) and 22.32 (dogs) respectively.

Below, we see a similar graph as the one above, but this one is of cats and dogs whose first intake type was “Stray”. We see a similar trend as before, but the percent prealtered is much lower than the chart above. The proportion of stray cats prealtered at intake dropped 3.61 percentage points from 2019 to 2023, and stray dogs dropped 8.7 percentage points.

Below shows the proportion of cats and dogs prealtered whose first intake type was “owner surrender”. The proportion of owner surrendered cats and dogs coming in prealtered is higher than stray cats and dogs. The sharpest drop is from 2019 to 2020, presumably an affect from COVID. Cats dropped 9.46 percentage points from 2019 to 2023 and dogs dropped 10.68 percentage points. Both owner surrendered cats and owner surrendered dogs experienced a larger percentage point drop from 2019 - 2023 than stray cats and stray dogs.

Percent Prelatered by Organization Type

Below shows a table of percent prealtered values. Minimum percent is the percent prealtered value from the year with the lowest percent, and maximum is from the year with the highest. For a majority of the organizations, 2019 (153/262) experienced the highest percent of animals coming in prealtered and 2023 (143/262) experienced the lowest. If one looks through the table, the gap between the minimum and maximum differ widely between organizations. This may indicate the presence of multiple factors affecting the percent of preatlered cats and dogs coming into the shelter.

Dog Sample

Similar to our first sample, our original Dog sample needed to be scrapped because a lot of entries were missing from 2024.

The Dog sample follows the same rules as our previous sample, where each observation references one animal - only dogs in this case - and their first intake listed on the software. The date range is the same, January 2019 to January 2023, and only organizations that have at least one entry each month were kept. This resulted in a sample size of 232 organizations.

Intake Counts by Year and Percent Change

A majority of the percent change from 2019 to 2023 values are negative (~77%). Of those that are negative, most of the percent change values range from -40% to -10%.

## [1] 256

Dog Analysis

Line Graphs

Below is a set of different line graphs where year is the x axis and y is the percent prealtered. Each line refers to a different group of dogs.

Intake Type

Since we kept the first intake date recorded for each animal, an intake type of “Return” does not make sense. However, there are some instances where an intake type of Return will show up. For example, the “Return” intake type most likely came from adoptions that occurred prior to January 2019.

Our sample shows that dogs transferred in have a similar prealtered proportion as dogs surrendered by an owner. Seized and strays have close percent prealtered rate as well.

Age Group at Intake

As expected, a higher proportion of older dogs come in prealtered vs younger dogs. Adult dogs experienced the largest drop in their prealtered proportion, with a drop of 13 percentage points.

Is it odd that the percent prealtered at intake for Senior and Adult dogs is not higher? I would expect that by the time an animal is a Senior or older Adult, that they would have been spayed or neutered at some point in their life.

## 
##      N      U      Y 
## 757236  62165 316465
## [1] 0.2785732

Dog Size

The large dogs dropped 14 percentage points from 2019 to 2023, and medium dogs dropped 11 percentage points. Small dogs dropped 9 percentage points. While all size groups dropped a similar amount (difference of 5 percentage points), small dogs have a much lower prealtered proportion than large and medium dogs.

When a shelter records the size of an animal, does age play any impact on that? For example, would a shelter organization label a weaned pitbull as small when it would be classified as large as an adult?

Sex

Male dogs have a higher proportion coming in prealtered than female dogs. Their proportions fell at similar rates.

LOS of Prealtered Dogs

Below is a line graph of the LOS for dogs by prealtered status. This includes all intakes and outcomes. As expected, dogs that arrive prealtered at intake have a lower LOS than dogs that arrive intact at intake. There is a difference of 1 from 2019 to 2021, and the different rises to 2 for 2022 and 2023.

LOS by Prealtered Status and Intake Type

Below is a line graph of the LOS by prealtered status at intake and intake type. All outcome types are included. In the legend, if you click on a group, it will remove that line from the graph. if you double click a group, it will remove all other lines

  • Owner/Guardian Surrender
    • The LOS for dogs prealtered at intake and dogs not prealtered at intake both increased, but the difference between the two remained the same except for 2020 and 2021.
  • Return
    • The LOS for dogs that were returned prealtered, did not change much from 2019 to 2023. It only increased by 1 in 2022 and 2023. The LOS for dogs that were returned not prealtered varied widely over the years. There was a sharp drop in 2020 and 2021, but then a big spike in 2022 and 2023.
  • Seized / Custody
    • Both prealtered and not prelatered seized dogs have a similar LOS trend, but there is a 2-3 day gap between them.
  • Stray
    • Stray dogs also have a similar trend between those that come in prealtered and those that come in intact. The difference between them is 1 day from 2019 to 2021, then it jumps to 2 days from 2022 to 2023. Prealtered stray dogs and prealtered seized dogs have the lowest LOS.
  • Transfer In
    • In 2023, both transferred in and returned dogs that came in intact had the longest LOS (13 days). From 2019 to 2023, the LOS for transferred in dogs that came in prealtered increased by 2 days while not prealtered transferred in dogs LOS increased by 4 days.

Only Adoption Outcome

Proportion by Regions for Dogs

The charts below show the percent of prealtered dogs that came into the shelter from 2019 to 2023 by region.

All of the regions have dropped in their percent of dogs coming in prealtered from 2019 to 2023. However, percentage point drop varies between the regions. For example, West North Central region had the largest percentage point drop at 18.23, and New England had the lowest percentage point drop at 2.6.

Cat Sample

The Cat sample follows the same rules as our the dogs sample. Each observation references cat and their first intake listed on the software. The date range is the same, January 2019 to January 2023, and only organizations that have at least one entry each month were kept. There are 227 organizations that met this criteria.

Intake Counts by Year and Percent Change

A majority of the percent change from 2019 to 2023 values are negative (~58%). Of those that are negative, most of the percent change values range from -20% to 0%.

## 
##    Adult Juvenile  Neonate   Senior   Weaned 
##   446590   280664   165844    63851   219350

Cat Analysis

Line Graphs

Below is a set of different line graphs where year is the x axis and y is the percent prealtered. Each line refers to a different group of cats.

Intake Type

Since we kept the first intake date recorded for each animal, an intake type of “Return” does not make sense. However, there are some instances where an intake type of Return will show up. For example, the “Return” intake type most likely came from adoptions that occurred prior to January 2019.

Our sample shows that cats transferred in have the highest proportion prealtered compared to the other intake types, beside “return”. There is a 12 percentage point gap between transfered in cats and owner surrendered that has remained constant since 2020. Stray and seized have the lowest proportion prealtered compared to the other intake types.

Age Group at Intake

As expected, a higher proportion of older cats come in prealtered vs younger cats. Both senior cats and adult cats experienced a drop of 7 percentages for percent prealtered from 2019 to 2024. Cats experience large gaps in the percent prealtered between seniors, adults, and juvenile aged cats. There is a 31 percentage point gap between senior and adult cats, and there is a 20 percentage point gap between adult and juvenile cats.

Same questions as for dogs, is it odd that the percent prealtered at intake for senior and adult dogs is not higher? I would expect that by the time an animal is a senior or older adult that they would have been spayed or neutered at some point in their life.

Sex

Male catss have a higher proportion coming in prealtered than female cats. Their proportions fell at similar rates.

LOS of Prealtered Cats

Below is a line graph of the LOS for cats by prealtered status. This includes all intakes and outcomes. As expected, cats that arrive prealtered at intake have a lower LOS than cats that arrive intact at intake. The gap in LOS between cats with a yes or no prealtered status is much larger than dogs, 6 days in 2019 compared to 1 day for dogs. The gap grew from 2019 to 2023, ending at a gap of 11 days.

LOS by Prealtered Status and Intake Type

Below is a line graph of the LOS by prealtered status at intake and intake type. All outcome types are included. In the legend, if you click on a group, it will remove that line from the graph. if you double click a group, it will remove all other lines

  • Owner/Guardian Surrender
    • The LOS for owner surrendered cats prealtered actually dropped slightly from 2019 to 2023. The LOS for cats not prealtered reamined steady from 2019 to 2022, then jumped up 2 percentage points in 2023.
  • Return
    • The trends for prealtered returned and cats and not prealtered were inverse of each other from 2019 to 2023. Non-prealtered cats experenced a large spike in 2022 and dropped to the lowest LOS of the 5 years.
  • Seized / Custody
    • Both prealtered and not prelatered seized cats have a similar LOS trend, but there is a 10-12 day gap between them.
  • Stray
    • Stray cats that came in prealtered have a consisitent LOS from 2019 to 2023, but not prealtered stray cats has progressiviley increased from 2019 to 2023. The LOS for not prealtered stray cats in 2023 is the highest of all the intake types at 26 days. This is a 7 percentage point increase from 2019 to 2023.
  • Transfer In
    • In 2019, the LOS for transferred in cats that came in not prealtered had the longest LOS of any other intake type, while transferred in cats that came in prealtered had the lowest LOS. They remained the lowest for the following years as well. Both dropped a similar amount in percentage points. 3 for prealtered cats and 4 for not prealtered cats.

Only Adoption Outcome

Proportion by Regions for Cats

The charts below show the percent of prealtered cats that came into the shelter from 2019 to 2023 by region.

All of the regions have dropped in their percent of cats coming in prealtered from 2019 to 2023. However, there are several regions that saw a minimal decrease in from 2019 to 2024. For example, 6 out of the 9 regions saw a drop of only 5 percentage points or less from 2019 to 2024. West South Central region saw the largest drop at 13.5 percentage points and West North Central region saw the smallest drop at 1.04 percentage points.

Statistical Tests

## 
##    Adult Juvenile  Neonate   Senior   Weaned 
##   703577   189823    36826   118861    86934
## [1] 155
## [1] 0.06131895
## [1] 0.05485814
## [1] 62320
## [1] 42262
## [1] "Chi Square Test Result"
## 
##  Pearson's Chi-squared test with Yates' continuity correction
## 
## data:  contingency_table
## X-squared = 15534, df = 1, p-value < 2.2e-16
## [1] 0.1549593
## Average Odds Ratio: 2.023944
## Average P-Value: 0.001947673
## Average 95% CI: [ 1.512351 , 2.70867 ]
## Strays have 2 times the odds of comming in not prealtered compared to owner surrenders when controlling for year

#Logistic Regression

Things to check prior to reporting results

  1. The outcome needs to be dichotomous. Pass
  2. The observations need to be independent/no duplicates. Unsure. Is there multiple intakes per animal? I can’t remember if I filtered for only unique animals. I am fairly certain I did. I removed duplicate unique animals, but there is always a possibility that a new id was given to the same animal. I do not this is the case enough to worry about it. Pass.
  3. No perfect predictor. Pass
  4. linearity of the logit. Not Applicable until I add continous variables
  5. Abscence of multi colinearity. Since my categorical variables are factors, they may be converted to 0/1 values which may cause there to be colinearity. PASS
  6. Adequate Sample Size. Pass
## 
##    Adult Juvenile   Senior 
##  1150167   470487   182712
## 
## Owner/Guardian Surrender                   Return                    Stray 
##                   463549                    17704                   630967
## 
## Call:
## glm(formula = prealtered ~ sex + size + intake_type + intake_year + 
##     `Age Group at Intake`, family = "binomial", data = dogs_dat_for_Log_r)
## 
## Coefficients:
##                                Estimate Std. Error  z value Pr(>|z|)    
## (Intercept)                   -0.039888   0.009295   -4.291 1.78e-05 ***
## sexM                           0.128371   0.006360   20.184  < 2e-16 ***
## sizeM                         -0.104564   0.007427  -14.079  < 2e-16 ***
## sizeS                         -0.306407   0.008099  -37.832  < 2e-16 ***
## intake_typeReturn              3.122264   0.043538   71.713  < 2e-16 ***
## intake_typeStray              -0.923166   0.006421 -143.777  < 2e-16 ***
## intake_year2020               -0.237768   0.009641  -24.663  < 2e-16 ***
## intake_year2021               -0.268477   0.009517  -28.209  < 2e-16 ***
## intake_year2022               -0.392334   0.009442  -41.551  < 2e-16 ***
## intake_year2023               -0.513205   0.009565  -53.654  < 2e-16 ***
## `Age Group at Intake`Juvenile -1.187520   0.011266 -105.406  < 2e-16 ***
## `Age Group at Intake`Senior    0.992228   0.008653  114.671  < 2e-16 ***
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## (Dispersion parameter for binomial family taken to be 1)
## 
##     Null deviance: 688803  on 566919  degrees of freedom
## Residual deviance: 610522  on 566908  degrees of freedom
## AIC: 610546
## 
## Number of Fisher Scoring iterations: 5
##                                       OR      2.5 %     97.5 %
## (Intercept)                    0.9608973  0.9435499  0.9785637
## sexM                           1.1369747  1.1228898  1.1512362
## sizeM                          0.9007175  0.8877015  0.9139245
## sizeS                          0.7360873  0.7244948  0.7478652
## intake_typeReturn             22.6977198 20.8411765 24.7196449
## intake_typeStray               0.3972592  0.3922911  0.4022901
## intake_year2020                0.7883857  0.7736289  0.8034241
## intake_year2021                0.7645434  0.7504139  0.7789389
## intake_year2022                0.6754782  0.6630926  0.6880952
## intake_year2023                0.5985743  0.5874573  0.6099017
## `Age Group at Intake`Juvenile  0.3049768  0.2983163  0.3117860
## `Age Group at Intake`Senior    2.6972385  2.6518809  2.7433718
## 
## Family: binomial 
## Link function: logit

## Likelihood ratio test
## 
## Model 1: prealtered ~ 1
## Model 2: prealtered ~ sex + intake_year + `Age Group at Intake` + ALL_QUIN
##   #Df  LogLik Df Chisq Pr(>Chisq)    
## 1   1 -344402                        
## 2  12 -322309 11 44185  < 2.2e-16 ***
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
##  Hosmer and Lemeshow goodness of fit (GOF) test
## 
## data:  y_num, fitted(dog_log_model)
## X-squared = 1228.7, df = 8, p-value < 2.2e-16
## 
## Call:
## glm(formula = prealtered ~ animal_type + `Age Group at Intake`, 
##     family = "binomial", data = dat_for_Log_R)
## 
## Coefficients:
##                                Estimate Std. Error z value Pr(>|z|)    
## (Intercept)                   -0.520355   0.003491 -149.08   <2e-16 ***
## animal_typeDog                -0.391409   0.004427  -88.41   <2e-16 ***
## `Age Group at Intake`Juvenile -1.193947   0.006073 -196.61   <2e-16 ***
## `Age Group at Intake`Senior    1.264550   0.006525  193.79   <2e-16 ***
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## (Dispersion parameter for binomial family taken to be 1)
## 
##     Null deviance: 1372299  on 1112219  degrees of freedom
## Residual deviance: 1267952  on 1112216  degrees of freedom
## AIC: 1267960
## 
## Number of Fisher Scoring iterations: 4
##                                      OR     2.5 %    97.5 %
## (Intercept)                   0.5943093 0.5902574 0.5983891
## animal_typeDog                0.6761037 0.6702626 0.6819958
## `Age Group at Intake`Juvenile 0.3030229 0.2994377 0.3066512
## `Age Group at Intake`Senior   3.5414970 3.4964908 3.5870824
##                           GVIF Df GVIF^(1/(2*Df))
## animal_type           1.086695  1        1.042447
## sex                   1.011217  1        1.005593
## intake_type           1.034467  2        1.008508
## intake_year           1.002735  4        1.000341
## `Age Group at Intake` 1.062738  2        1.015328
## ALL_QUIN              1.021425  4        1.002653
## 
## Call:
## glm(formula = prealtered ~ animal_type + sex + intake_type + 
##     intake_year + `Age Group at Intake` + ALL_QUIN, family = "binomial", 
##     data = dat_for_Log_R)
## 
## Coefficients:
##                                Estimate Std. Error  z value Pr(>|z|)    
## (Intercept)                    0.138988   0.007354   18.900  < 2e-16 ***
## animal_typeDog                -0.274313   0.004673  -58.702  < 2e-16 ***
## sexM                           0.182303   0.004514   40.384  < 2e-16 ***
## intake_typeReturn              3.247897   0.034155   95.092  < 2e-16 ***
## intake_typeStray              -0.812477   0.004567 -177.908  < 2e-16 ***
## intake_year2020               -0.189683   0.006937  -27.345  < 2e-16 ***
## intake_year2021               -0.227364   0.006834  -33.272  < 2e-16 ***
## intake_year2022               -0.324021   0.006785  -47.755  < 2e-16 ***
## intake_year2023               -0.441306   0.006888  -64.073  < 2e-16 ***
## `Age Group at Intake`Juvenile -1.224416   0.006260 -195.582  < 2e-16 ***
## `Age Group at Intake`Senior    1.193903   0.006791  175.809  < 2e-16 ***
## ALL_QUIN2                     -0.322130   0.007663  -42.035  < 2e-16 ***
## ALL_QUIN3                      0.056703   0.007031    8.064 7.36e-16 ***
## ALL_QUIN4                     -0.186894   0.006752  -27.679  < 2e-16 ***
## ALL_QUIN5                     -0.429156   0.006652  -64.512  < 2e-16 ***
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## (Dispersion parameter for binomial family taken to be 1)
## 
##     Null deviance: 1372299  on 1112219  degrees of freedom
## Residual deviance: 1191871  on 1112205  degrees of freedom
## AIC: 1191901
## 
## Number of Fisher Scoring iterations: 5
##                                       OR      2.5 %     97.5 %
## (Intercept)                    1.1491101  1.1326664  1.1657926
## animal_typeDog                 0.7600943  0.7531646  0.7670879
## sexM                           1.1999775  1.1894072  1.2106418
## intake_typeReturn             25.7361486 24.0696961 27.5179772
## intake_typeStray               0.4437577  0.4398034  0.4477475
## intake_year2020                0.8272212  0.8160507  0.8385447
## intake_year2021                0.7966307  0.7860322  0.8073722
## intake_year2022                0.7232352  0.7136810  0.7329172
## intake_year2023                0.6431958  0.6345714  0.6519374
## `Age Group at Intake`Juvenile  0.2939293  0.2903448  0.2975580
## `Age Group at Intake`Senior    3.2999342  3.2563032  3.3441497
## ALL_QUIN2                      0.7246043  0.7138021  0.7355701
## ALL_QUIN3                      1.0583417  1.0438566  1.0730278
## ALL_QUIN4                      0.8295319  0.8186261  0.8405831
## ALL_QUIN5                      0.6510581  0.6426244  0.6596024
##                                    2.5 %     97.5 %
## (Intercept)                    1.1326664  1.1657926
## animal_typeDog                 0.7531646  0.7670879
## sexM                           1.1894072  1.2106418
## intake_typeReturn             24.0696961 27.5179772
## intake_typeStray               0.4398034  0.4477475
## intake_year2020                0.8160507  0.8385447
## intake_year2021                0.7860322  0.8073722
## intake_year2022                0.7136810  0.7329172
## intake_year2023                0.6345714  0.6519374
## `Age Group at Intake`Juvenile  0.2903448  0.2975580
## `Age Group at Intake`Senior    3.2563032  3.3441497
## ALL_QUIN2                      0.7138021  0.7355701
## ALL_QUIN3                      1.0438566  1.0730278
## ALL_QUIN4                      0.8186261  0.8405831
## ALL_QUIN5                      0.6426244  0.6596024